The purpose of this study is to evaluate the effect of a peer comparison feedback report, combined with a best practices toolkit, on the volume of antimicrobial use by general medicine physicians. The study also aims to understand the qualitative and quantitative impact of peer comparison feedback on antimicrobial prescribing in hospital.

This study will leverage data from GEMINI, a hospital research collaborative that collects administrative and clinical data from 30+ Ontario hospitals, to create these peer comparison feedback reports.

Antimicrobial resistance (AMR) is a critical threat to human health, having led to the deaths of 14,000 people in Canada in 2018 and 1.25 million people globally in 2019. There has been limited success in overcoming AMR through developing new antimicrobials, primarily due to inadequate investment in new antimicrobial discovery, and economic and regulatory hurdles. Acquired resistance is also likely to occur as newer agents come to market. The World Health Organization and the Public Health Agency of Canada have identified overuse of antimicrobials as a primary ongoing driver of AMR. Therefore, optimizing antimicrobial use must be a cornerstone of our response to the AMR crisis.

Antimicrobial stewardship (AMS) interventions (activities designed to safely improve antimicrobial use) can lead to reductions in antimicrobial use, antimicrobial costs, hospital length of stay, and C. difficile rates. Previous work has demonstrated that audit and feedback with peer comparison in a primary care setting can lead to a clinically significant reduction in antimicrobial use, with no demonstrable safety concerns. For example, the study team previously tested a peer comparison letter to primary care providers in the highest quartile of antimicrobial prescribing across Ontario and found a clinically significant 4% relative reduction in total antimicrobial prescribing. This extrapolates to nearly 150,000 fewer prescriptions and drug-cost savings of $2.7 million annually for the province.

General medicine wards are an important target for antimicrobial stewardship efforts, as they care for the largest group of hospital inpatients, patients are often medically complex, and admissions are heterogeneous in condition/disease. However, physician-level prescribing data for hospitalized patients has historically been difficult to collect and attribute to a specific clinician, limiting rigorous evaluation. This trial will leverage GEMINI's pharmacy data to create physician-level peer-comparison reports for physicians practicing on general medical wards. Building on a previous CIHR-funded RCT led by Drs. Desveaux and Ivers, this trial will also develop a mixed methods process evaluation methodology to assess fidelity and mechanism of action.

Inclusion Criteria:

1. Patient-level: All adult inpatients (>=18 years of age) discharged within the baseline evaluation period and follow up period will be included in a physician report if they match the following criteria: a) Total in-hospital length of stay (LOS) is less than or equal to 14 days; b) and admitted to or discharged from a general medicine or hospitalist ward; c) admitted via the emergency department.
2. Physician-level: Physicians will be included if they have at least 50 hospitalizations attributed to them during the baseline period. Physicians who practice at multiple sites will only be included at the hospital with more encounters.
3. Hospital-level: In total, 29 teaching and community hospitals that provide data to GEMINI, with comprehensive geographic coverage of Ontario, will be included in this trial.

Exclusion Criteria:

1. Patient-level: All data from ICU dates of care will be excluded from the trial (defined by ICU admission start and stop dates). All antimicrobial use data from ICU dates will be censored and excluded from attribution to GM physicians. Additionally, all patients with ICD-10-CA code Z51.5 for palliative care will be excluded owing to expected differences in antimicrobial management strategies.
2. Physician-level: No additional exclusion criteria will be used.
3. Hospital-level: Two hospitals in the GEMINI network will be excluded from this trial due to their lack of GM wards (a cancer hospital and a COVID-19 reactivation hospital). Additionally, any hospitals in the GEMINI network that do not meet annual data quality standards for antimicrobial use or other data fields will not be included in the trial.

Individual participant data for this trial will be collected passively through GEMINI's existing data collection procedures. Therefore, the data being leveraged for this trial (i.e., a research-ready dataset of all inpatient antimicrobial prescribing across 60% of Ontario's medical hospital beds) will be available for other researchers to use as soon as data validation is complete. GEMINI's data dictionary, which will contain further information about the data used for this trial, is available here: https://geminimedicine.ca/the-gemini-database/.

To use GEMINI data, researchers must follow GEMINI's data access process, which can be found here: https://geminimedicine.ca/access-gemini-data/.

Beginning in September 2025, with no end date.

GEMINI data can be accessed through GEMINI's secure research environment, once approved by GEMINI's Projects & Publications Committee and the appropriate REB Office. To access data, researchers must submit a project proposal to GEMINI.Research@unityhealth.to, using the standardized template available here: https://geminimedicine.ca/access-gemini-data/